from sklearn.model_selection import StratifiedKFold
from evalml.preprocessing.data_splitters.balanced_classification_sampler import (
BalancedClassificationSampler
)
from evalml.preprocessing.data_splitters.base_splitters import (
BaseUnderSamplingSplitter
)
from evalml.preprocessing.data_splitters.training_validation_split import (
TrainingValidationSplit
)
[docs]class BalancedClassificationDataTVSplit(BaseUnderSamplingSplitter):
"""Data splitter for generating training and validation split using Balanced Classification Data Sampler."""
[docs] def __init__(self, balanced_ratio=4, min_samples=100, min_percentage=0.1, test_size=0.25, shuffle=True, random_seed=0):
"""Create Balanced Classification Data TV splitter
Arguments:
balanced_ratio (float): The largest majority:minority ratio that is accepted as 'balanced'. For instance, a 4:1 ratio would be
represented as 4, while a 6:5 ratio is 1.2. Must be greater than or equal to 1 (or 1:1). Defaults to 4.
min_samples (int): The minimum number of samples that we must have for any class, pre or post sampling. If a class must be downsampled, it will not be downsampled past this value.
To determine severe imbalance, the minority class must occur less often than this and must have a class ratio below min_percentage.
Must be greater than 0. Defaults to 100.
min_percentage (float): The minimum percentage of the minimum class to total dataset that we tolerate, as long as it is above min_samples.
If min_percentage and min_samples are not met, treat this as severely imbalanced, and we will not resample the data.
Must be between 0 and 0.5, inclusive. Defaults to 0.1.
test_size (float): The size of the test split. Defaults to 0.25.
shuffle (bool): Whether or not to shuffle the data before splitting. Defaults to True.
random_seed (int): The seed to use for random sampling. Defaults to 0.
"""
self.sampler = BalancedClassificationSampler(balanced_ratio=balanced_ratio, min_samples=min_samples, min_percentage=min_percentage, random_seed=random_seed)
super().__init__(sampler=self.sampler, n_splits=1, random_seed=random_seed)
self.shuffle = shuffle
self.test_size = test_size
self.splitter = TrainingValidationSplit(test_size=test_size, shuffle=shuffle, random_state=random_seed)
[docs]class BalancedClassificationDataCVSplit(BaseUnderSamplingSplitter):
"""Data splitter for generating k-fold cross-validation split using Balanced Classification Data Sampler."""
[docs] def __init__(self, balanced_ratio=4, min_samples=100, min_percentage=0.1, n_splits=3, shuffle=True, random_seed=0):
"""Create Balanced Classification Data CV splitter
Arguments:
balanced_ratio (float): The largest majority:minority ratio that is accepted as 'balanced'. For instance, a 4:1 ratio would be
represented as 4, while a 6:5 ratio is 1.2. Must be greater than or equal to 1 (or 1:1). Defaults to 4.
min_samples (int): The minimum number of samples that we must have for any class, pre or post sampling. If a class must be downsampled, it will not be downsampled past this value.
To determine severe imbalance, the minority class must occur less often than this and must have a class ratio below min_percentage.
Must be greater than 0. Defaults to 100.
min_percentage (float): The minimum percentage of the minimum class to total dataset that we tolerate, as long as it is above min_samples.
If min_percentage and min_samples are not met, treat this as severely imbalanced, and we will not resample the data.
Must be between 0 and 0.5, inclusive. Defaults to 0.1.
n_splits (int): The number of splits to use for cross validation. Defaults to 3.
shuffle (bool): Whether or not to shuffle the data before splitting. Defaults to True.
random_seed (int): The seed to use for random sampling. Defaults to 0.
"""
self.sampler = BalancedClassificationSampler(balanced_ratio=balanced_ratio, min_samples=min_samples, min_percentage=min_percentage, random_seed=random_seed)
super().__init__(sampler=self.sampler, n_splits=n_splits, random_seed=random_seed)
self.shuffle = shuffle
self.splitter = StratifiedKFold(n_splits=n_splits, shuffle=shuffle, random_state=random_seed)